some bitter truths about the instability of reinforcement learning. h/t @hardmaruhttps://twitter.com/hardmaru/status/983370851436322816 …
-
@demishassabis should read this since he seems to believe that the brain uses backprop driven by RL. Pathetic but Marcus has nothing better to offer. He wants everyone to go back to GOFAI, not realizing that mainstream AI never got rid of GOFAI even if they feign otherwise.
-
-
Not really since deep learning is still GOFAI, loud denials notwithstanding. A label is a symbol, which has no meaning. This is why face recognizers do not know what face means and GO players do not know what playing means. Unless you got a replacement for DL, you got nothing.
-
I take it that, by "hybrid", you mean a mixture of symbolic and non-symbolic. Any symbolic approach to AGI will fail. The late Hubert Dreyfus understood and explained why representationalism was nonsense but mainstream AI is pathologically hard of hearing.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.